ABSENCE OF FAST-MOVING IRON IN AN INTERMEDIATE TYPE Ia SUPERNOVA BETWEEN NORMAL AND SUPER-CHANDRASEKHAR

In order to determine the explosion date of iPTF13asv, we follow Nugent et al. (2011) and model the early PTF R-band light curve of iPTF13asv as a freely expanding fireball where the luminosity increases as ∝t² and the temperature remains constant. Restricting ourselves to the light curve within four days of discovery, we find a best fit (the inset in Figure 1) at an explosion date of April 29.4 ± 0.3 (95% confidence interval) with a fitting ${\chi }^{2}=4.3$ for five degrees of freedom. The best-fit light curve is also consistent with the non-detection upper limit on April 30.4.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

If we generalize the t² model with a power-law model, we obtain strong degeneracy between the explosion date and the power-law index over a large range of the parameter space. In fact, Firth et al. (2015) analyzed the rise behavior of a large sample of SNe Ia with the power-law model and also found that the power-law indices have large uncertainties with a mean value of 2.5. This is probably because shallow-deposited ⁵⁶Ni heats up SN photospheres. Furthermore, there could be a dark time between the SN explosion and the SN light curve powered by radioactive decay on the diffusive timescale for the shallowest deposition of ⁵⁶Ni in the ejecta (Piro 2012; Piro & Morozova 2015). Hence, it is nontrivial to estimate the exact explosion date purely from the early light curve. Since the following analysis and discussion are not very sensitive to the exact explosion date, for simplicity, we adopt the explosion date determined by the t² model.

The most striking feature of these early-phase spectra is the absence of iron absorption. We used SYN++ (Thomas et al. 2011b) to perform spectral feature identification on the spectrum taken 11 days after explosion (or equivalently, −13.7 days with respect to the B-band maximum which is determined in Section 3.3). As highlighted in gray in Figure 3, the spectrum shows no signature of either Fe ii or Fe iii.

Zoom In Zoom Out Reset image size

We also compare early-phase spectra of iPTF13asv to those of well-studied normal and overluminous type Ia events in the top panels of Figure 4. Most spectra in comparison, including the spectrum of the super-Chandrasekhar SN2009dc at −7 days, clearly show the existence of iron absorption features. The exceptions are two super-Chandrasekhar events, SN2006gz and SN2007if, which have weak or no iron absorption.

Zoom In Zoom Out Reset image size

As a result of nucleosynthesis and mixing during SN explosions, iron commonly manifests itself as absorption features in SN spectra, either as Fe ii at low effective temperatures or as Fe iii at higher temperatures. The absence of iron at early phases implies that weak mixing during the SN explosion confines synthesized iron group elements in low-velocity regions of the ejecta.

The centric concentration of iron can be verified by strong UV emission at the same time, as the iron group elements are the main absorbers of photons below 3500 Å. However, we did not trigger Swift observations at early phases because the SN is located beyond our trigger criteria of 100 Mpc. In comparison to the spectral shape of SN2011fe, the spectral shape of iPTF13asv at −7 days (top right panel of Figure 4) indicates stronger fluxes at shorter wavelengths, hinting a strong emission in the UV.

The weak mixing of iron in the SN explosion may have strong implications for the explosion mechanism and will be discussed in Section 5.3. As shown in the bottom two panels of Figure 4, iron features appear in the iPTF13asv spectra around and after maximum. In the next few subsections, we investigate the specifics of iPTF13asv.

In order to determine the light curve shape parameters of iPTF13asv, we use the SALT2 software (Guy et al. 2007) to fit its optical light curve (see Figure 1 for the SALT2 best-fit light curves). The best-fit light curve gives a rest-frame B-band peak magnitude ${m}_{B}=16.28\pm 0.03$ on May 18.12 ± 0.09. We set this B-band peak date as t = 0 in the rest of this paper. The fit also gives a color term $c=-0.16\pm 0.02$ and two shape parameters ${x}_{0}=0.0055\pm 0.0001$ and ${x}_{1}=0.37\pm 0.09$. Based on the fitted SALT2 light curve, we derive a color ${(B-V)}_{0}=-0.14\pm 0.03$ at the B-band maximum. We also obtain ${\rm{\Delta }}{m}_{15}=1.03\pm 0.01$ from x₁ by using the relation in Guy et al. (2007).

The local extinction in the host galaxy of iPTF13asv is probably minor for several reasons. First, Figure 5 compares the B − V colors between iPTF13asv (${\rm{\Delta }}{m}_{15}=1.03\;{\rm{mag}}$) and SN2011fe (${\rm{\Delta }}{m}_{15}=1.10\;{\rm{mag}}$). Since SN2011fe is unreddened by its host (Nugent et al. 2011; Vinkó et al. 2012), similar colors around maximum suggest that iPTF13asv also has little local extinction. After the maximum, iPTF13asv has a slightly blue color compared to SN2011fe, probably due to the different stretch of these two events (Nobili & Goobar 2008), until they join the Lira relation after +30 days. Second, the intrinsic color $B-V=0.95$ mag of iPTF13asv at +35 days is consistent with the latest calibration of the Lira relation (Burns et al. 2014). Third, the absence of Na i D absorption in the low-resolution optical spectra also implies weak extinction in the host galaxy. Given a typical velocity dispersion of $10\;{\rm{km}}\;{{\rm{s}}}^{-1}$ for a dwarf galaxy (see Section 3.5 for a discussion of the iPTF13asv host galaxy; Walker et al. 2007), we derive from the highest signal-to-noise ratio spectrum that the equivalent width for each of the Na i D lines is less than $0.2\;$Å(5-σ). Using the empirical relation in Poznanski et al. (2012), we find that the extinction $E(B-V)\lt 0.06$. Therefore, in what follows, we neglect the local extinction correction.

Zoom In Zoom Out Reset image size

After correction for Galactic extinction we derive an absolute peak magnitude of iPTF13asv in its rest-frame B band to be −19.84 ± 0.06. This is about 0.5 mag brighter than normal SNe Ia at peak.

The k-correction in the UV and optical wavelengths is negligible. Synthetic photometry using both the Nugent SN Ia template (Nugent et al. 2002) and the observed HST UV spectra of SN2011fe (Mazzali et al. 2014) shows that the k-correction is less than 0.1 mag in the optical and 0.2 mag in the Swift/UVOT UV filters.

In Figure 6, the optical light curves of iPTF13asv are compared to those of well-studied SNe, including normal SN2011fe; overluminous SN1999aa and SN1991T; and super-Chandrasekhar SN2006gz, SN2007if, and SN2009dc. All the light curves have been offset to match their peak magnitudes and the epoch of the B-band maxima. Figure 6 illustrates that (1) the light curve width of iPTF13asv is similar to those of normal events and narrower than those of overluminous and super-Chandrasekhar events, except for SN2006gz, (2) iPTF13asv shows an isolated secondary maximum in the I-band whose strength is weaker than those observed in SN2011fe and SN1991T, and (3) iPTF13asv matches well to the super-Chandrasekhar SN2006gz in the B- and V-band light curves, but SN2006gz has a much stronger near-IR secondary peak.

Zoom In Zoom Out Reset image size

Figure 7 compares iPTF13asv in the Swift/UVOT uvm2 and uvw2 filters to a large sample of normal, overluminous, and super-Chandrasekhar SNe Ia observed by Swift (Milne et al. 2013; Brown 2014). While the uvw2 filter has a non-negligible leakage in long wavelengths, the uvm2 filter does not have a significant leakage and therefore provides the best available measurements of the UV flux. The figure shows that, like super-Chandrasekhar events, iPTF13asv is more luminous in the UV than the majority of normal events. Furthermore, Milne et al. (2013) divided SNe Ia into different subclasses based on their Swift/UVOT colors. We cannot make a direct comparison here because only uvm2 and uvw2 data are available for iPTF13asv. An indirect comparison is that iPTF13asv is brighter than SN2011fe by half a magnitude in the optical and by $\simeq 0.7$ mag in the UV. Since SN2011fe belongs to the NUV-blue subclass (Brown et al. 2012), iPTF13asv probably belongs to the same subclass.

Zoom In Zoom Out Reset image size

We also compare the IR light curves of iPTF13asv in the J- and H-band to the most recent light curve template for normal Type Ia events (Figure 1; Stanishev et al. 2015) and find that the sparsely sampled light curves of iPTF13asv roughly follow the template. The peak magnitudes of iPTF13asv are ${M}_{J}=-18.83\pm 0.08$ and ${M}_{H}=-18.16\pm 0.19$, compared to the median peak magnitudes of ${M}_{J}=-18.39$ and ${M}_{H}=-18.36$ with rms scatters of ${\sigma }_{J}=0.116\pm 0.027$ and ${\sigma }_{H}=0.085\pm 0.16$ for normal SNe Ia (Barone-Nugent et al. 2012). The secondary maximum of iPTF13asv is clearly seen in the J-band and H-band light curves as well as the I-band light curve, indicating concentration of iron group elements in the central region of the ejecta (Kasen 2006). This is in accordance with the absence of iron in the outer ejecta.

3.4.1. Spectral Cross-matching

We use the latest versions of both SN Identification (SNID; Blondin & Tonry 2007) and SuperFit (Howell et al. 2005) to spectroscopically classify iPTF13asv. Before −5 days, not surprisingly, neither tools find a good match for iPTF13asv spectra partly because they do not have many early-phase SNe in their templates and partly because the iron absorption is absent in the early-phase spectra of iPTF13asv. Around and after maximum, based on the first 5 best matches, both SNID and SuperFit find that iPTF13asv spectroscopically resembles normal SNe Ia (Table 3).

Table 3.  SNID Results

Phase	First Five Best Matchesa
−6.8	05eu@−5.2 (normal)	03ic@−4.1 (normal)	08Z@−4.3 (normal)	05na@−1.5 (normal)	06cc@−9.7 (normal)
0.0	96ai@+2.2 (normal)	07F@+3.0 (normal)	94ae@0.0 (normal)	03cg@−2.1 (normal)	94ae@+0.9 (normal)
$+6.8$	08Z@+7.6 (normal)	05na@+3.4 (normal)	99aa@+1.8 (peculiar)	01fe@+6.2 (normal)	06cz@−2.0 (91T-like)
$+13.3$	08Z@+12.3 (normal)	07ca@+13.3 (normal)	07bj@+12.0 (normal)	03fa@+13.4 (91T-like)	03kf@+13.4 (normal)

Note.

^aThe format in this column is name@phase (subclass).

Download table as:  ASCIITypeset image

3.4.2. Spectral Comparison to Well-studied SNe

3.4.3. Si ii Velocities

We further measure the expansion velocity evolution of iPTF13asv by fitting a Gaussian kernel to the Si ii 6355 line in each spectrum. The continuum is modeled by a linear regression to regions at both sides of the line. Then we fit a linear model to the velocity measurements between −10 and +10 days and estimate a velocity of $(1.0\pm 0.1)\times {10}^{4}\;{\rm{km}}\;{{\rm{s}}}^{-1}$ and a velocity gradient close to zero at peak.

Figure 8 shows Si ii velocities at maximum versus peak magnitudes. As can be seen in the figure, iPTF13asv has a Si ii velocity lower than the majority of normal SNe Ia and similar to super-Chandrasekhar events. Figure 9 compares Si ii velocity gradients at maximum versus peak magnitudes. Again, like super-Chandrasekhar events, iPTF13asv has a velocity gradient close to zero, lower than the majority of normal events.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.4.4. Carbon Signatures

We also note that iPTF13asv shows weak but persistent C ii absorption features until at least a week after maximum. Figure 10 shows SYN++ fits to iPTF13asv spectra of high signal-to-noise ratios, demonstrating the existence of C ii 6580 and C ii 7234 lines. The velocities of these C ii lines evolve from $\simeq {\rm{14,000}}\;{\rm{km}}\;{{\rm{s}}}^{-1}$ at −13.7 days to $\simeq {\rm{11,000}}\;{\rm{km}}\;{{\rm{s}}}^{-1}$ at $+6.3$ days.

Zoom In Zoom Out Reset image size

About 30% of normal SNe are estimated to reveal the C ii 6580 and C ii 7234 absorption notches in early phases (Parrent et al. 2011; Thomas et al. 2011a; Silverman & Filippenko 2012). These C ii features usually disappear before maximum. In contrast, some super-Chandrasekhar events show strong and persistent C ii features even after maximum. Figure 11 compares the spectra of iPTF13asv at one week after maximum to those of well-studied SNe at similar phases. As can be seen, neither SN1991T nor SN1999aa has the carbon feature at this phase; the carbon signature of iPTF13asv is not as strong as those seen in the super-Chandrasekhar SN2009dc.

Zoom In Zoom Out Reset image size

After iPTF13asv faded away, we obtained a low signal-to-noise ratio spectrum of its apparent host galaxy SDSS J162254.02+185733.8. The spectrum only shows Hα emission at the redshift of iPTF13asv. We fit a Gaussian profile to the Hα line and measure a luminosity of $3\times {10}^{38}\;\mathrm{erg}\;{{\rm{s}}}^{-1}$. We adopt the empirical relation between Hα luminosity and star formation rate (Kennicutt 1998) and obtain a star formation rate of $2\times {10}^{-3}\;{M}_{\odot }\;{{\rm{yr}}}^{-1}$ for the host galaxy.

Next, we construct the spectral energy distribution (SED) of the host galaxy with optical photometry from SDSS and near-IR photometry measured on the SN reference images. The SED is then modeled with a galaxy synthesis code called the Fitting and Assessment of Synthetic Templates (Kriek et al. 2009) assuming an exponentially decaying star formation history and a solar metallicity. The best-fit model gives a galaxy age of ${10}^{8.6}$ years and a stellar mass ${\mathrm{log}}_{10}({M}_{{\rm{stellar}}}/{M}_{\odot })={7.85}_{-0.4}^{+0.5}$ with a reduced ${\chi }^{2}=1.6$ (Figure 12). The best-fit model also shows no ongoing star-forming activity. Because SED fitting models are usually insensitive to very low star-forming rates, the best-fit model is consistent with a low star formation rate derived from the Hα flux. The derived star formation rate and the stellar mass of the iPTF13asv host galaxy follow the empirical relation between stellar mass and star formation rate (Foster et al. 2012).

Zoom In Zoom Out Reset image size

Since the host galaxy spectrum does not show [N ii] lines, we estimate an upper limit of $\mathrm{log}($[N ii 6548/Hα]$)\lt -0.87$. Using Denicoló et al. (2002), we derived a metallicity upper limit of $12+\mathrm{log}({\rm{O}}/{\rm{H}})\lt 8.3$. In fact, using the mass–metallicity relation (Foster et al. 2012), we estimate a gas-phase metallicity of $12+\mathrm{log}({\rm{O/H}})\sim 8$ for the host galaxy. Compared to the host galaxy samples of SNe Ia in Pan et al. (2014) and Wolf et al. (2016), SDSS J162254.02+185733.8 is one of the least massive and most metal-poor galaxies that host SNe Ia.

Given the wavelength coverage of the iPTF13asv spectra, we first construct a pseudo-bolometric light curve between 3500 and 9700 Å. In order to calibrate the absolute fluxes of these spectra, we use interpolated optical light curves to "warp" the spectra. Then the spectra are integrated to derive the pseudo-bolometric light curve.

Due to the sparsely sampled UV and IR light curves, it is difficult to estimate the UV and IR radiation at different phases. Therefore we calculate optical-to-bolometric correction factors with a spectral template (Hsiao et al. 2007). In this calculation, we find that the UV correction reaches about 25% before the B-band maximum and quickly drops to less than 5% around and after the B-band maximum. Given the inference that the SN might be UV-luminous before maximum and the observational fact that the SN is among the UV-bright SNe Ia around maximum, our calculated correction probably underestimates the UV radiation. Using the Swift data around maximum, we estimate that this UV correction introduces a systematic uncertainty of a few percent to the bolometric luminosity. Around maximum when the SN cools down, the UV contribution to the bolometric luminosity becomes even less important.

In the IR, the correction above 9700 Å is below 10% around the B-band maximum, then reaches a maximum of 24% around the secondary maximum in the near-IR. At the epochs with IR data, we find that the calculated correction is consistent with the IR measurements.

The final bolometric light curve is shown in Figure 13. We further employ Gaussian process regression to derive a maximum bolometric luminosity ${L}_{\mathrm{max}}=(2.2\pm 0.2)\;\times {10}^{43}\;{\rm{erg}}\;{{\rm{s}}}^{-1}$ at −2.6 days.

Zoom In Zoom Out Reset image size

Next, we follow the procedure in Scalzo et al. (2012, 2014a) to derive the ⁵⁶Ni mass and the total ejecta mass. First, the ⁵⁶Ni mass can be estimated through the following equation

$$\begin{eqnarray}&&{L}_{\mathrm{max}}=\alpha S({t}_{R}),\end{eqnarray} \tag{ 1 }$$

where $S({t}_{R})$ is the instantaneous radioactive power at the bolometric luminosity maximum. α is an efficiency factor of order unity, depending on the distribution of ⁵⁶Ni (Jeffery et al. 2006). We adopt a fiducial value of $\alpha =1.3$ following Scalzo et al. (2012). The radioactive power of ${}^{56}{\rm{Ni}}{\to }^{56}\mathrm{Co}\to {}^{56}\mathrm{Fe}$ is (Nadyozhin 1994)

$$\begin{eqnarray}&&S({t}_{r})=[6.31\mathrm{exp}(-{t}_{r}/8.8)+1.43\mathrm{exp}(-{t}_{r}/111)]{M}_{{\rm{Ni}}},\end{eqnarray} \tag{ 2 }$$

where S(t) is in units of ${10}^{43}\;{\rm{erg}}\;{{\rm{s}}}^{-1}$ and ${M}_{{\rm{Ni}}}$ is in units of ${M}_{\odot }$. With the measured maximum bolometric luminosity ${L}_{\mathrm{max}}=(2.2\pm 0.2)\times {10}^{43}\;{\rm{erg}}\;{{\rm{s}}}^{-1}$ at −2.6 days, we estimate a ⁵⁶Ni mass of $(0.77\pm 0.07){M}_{\odot }$.

About one month after the SN maximum, the SN debris expands approximately in a homologous manner. At this time, most ⁵⁶Ni atoms have decayed to ⁵⁶Co. Hence the total luminosity can be approximated by

$$\begin{eqnarray}&&L(t)=[1-\mathrm{exp}(-{({t}_{0}/t)}^{2})]{S}_{\gamma }(t)+{S}_{{e}^{+}}(t),\end{eqnarray} \tag{ 3 }$$

where S_γ and ${S}_{{e}^{+}}$ are the decay energy of ⁵⁶Co carried by γ-ray photons and positrons. At time t₀, the mean optical path of γ-ray photons becomes unity. For a given density and velocity profile, t₀ reflects the column density along the line of sight. We fit Equation (3) to the bolometric light curve of iPTF13asv after +20 days and obtained ${t}_{0}=44.2\pm 2.0\;{\rm{days}}$.

Next, we estimate the total ejecta mass of iPTF13asv. If we assume a density profile $\rho (v)\propto \mathrm{exp}(-v/{v}_{e})$ where v_e is a scale velocity, then the ejecta mass can be expressed as

$$\begin{eqnarray}&&{M}_{\mathrm{ej}}=\displaystyle \frac{8\pi }{{\kappa }_{\gamma }q}{({v}_{e}{t}_{0})}^{2},\end{eqnarray} \tag{ 4 }$$

where ${\kappa }_{\gamma }$ is the Compton scattering opacity for γ-ray photons. The value of ${\kappa }_{\gamma }$ is expected to lie in the range between 0.025 and $0.033\;{{\rm{cm}}}^{2}\;{{\rm{g}}}^{-1}$ (Swartz et al. 1995). We adopt a value of $0.025\;{{\rm{cm}}}^{2}\;{{\rm{g}}}^{-1}$ for the optically thin regime. The form factor q describes the distribution of ⁵⁶Ni and thus ⁵⁶Co (Jeffery 1999). For evenly mixed ⁵⁶Ni, the value of q is close to one-third. Taking element stratification and mixing in the interfaces into account, Scalzo et al. (2014a) found that $q=0.45\pm 0.05$. Here, we adopt q = 0.45 in our estimation.

The value of v_e can be obtained by conservation of energy. The total kinetic energy of the ejecta is $6{M}_{\mathrm{ej}}{v}_{e}^{2}$. Neglecting the radiation energy, the total kinetic energy is equal to the difference between the nuclear energy released in the explosion and the binding energy of the exploding WD. The binding energy of a rotating WD with mass M_ej and central density ${\rho }_{c}$ is given in Yoon & Langer (2005). Here we restrict the central density to lie between 10⁷ and ${10}^{10}\;{\rm{g}}\;{{\rm{cm}}}^{-3}$.

If we further assume that the ejecta is composed of unburned CO and synthesized Si, Fe, and Ni, then the nuclear energy of the SN explosion is formulated in Maeda & Iwamoto (2009) as a function of mass M_ej and mass fractions ${f}_{{\rm{Fe}}}$, ${f}_{{\rm{Ni}}}$, and ${f}_{{\rm{Si}}}$. The ratio $\eta ={f}_{{\rm{Ni}}}/({f}_{{\rm{Ni}}}+{f}_{{\rm{Fe}}})$ is also a function of ${\rho }_{c}$. Following Scalzo et al. (2014a), we adopt a Gaussian prior

$$\begin{eqnarray}&&\eta =0.95-0.05{\rho }_{c,9}\pm 0.03\mathrm{max}(1,{\rho }_{c,9}),\end{eqnarray} \tag{ 5 }$$

where ${\rho }_{c,9}$ is ${\rho }_{c}$ in units of ${10}^{9}\;{\rm{g}}\;{{\rm{cm}}}^{-3}$. In addition, we restrict the mass fraction ${f}_{{\rm{CO}}}$ less than 10%.

Based on the above assumptions, with a given set of ejecta mass M_ej, central density ${\rho }_{c}$, and the mass fractions of different elements, we can calculate the maximum bolometric luminosity and t₀ in Equation (3) and compare them with our measurements of iPTF13asv. Here we perform Markov-Chain Monte Carlo simulations for one million steps and obtain ${M}_{\mathrm{Ni}}={0.81}_{-0.18}^{+0.10}$ and ${M}_{\mathrm{ej}}={1.44}_{-0.12}^{+0.44}{M}_{\odot }$ at a 95% confidence level.

In order to explain the almost constant Si ii velocity in super-Chandrasekhar events, Scalzo et al. (2010, 2012) hypothesize a stationary shell detached from the ejecta. The shell is accelerated to a constant speed v_sh by colliding with fast-moving ejecta with velocities greater than v_sh. In fact, some simulations of WD mergers show that the outermost material forms such a stationary envelope that collides with fast-moving ejecta (Hoeflich & Khokhlov 1996). Following the calculation procedure in Scalzo et al. (2010, 2012), we derive an envelope mass of ${0.15}_{-0.01}^{+0.11}{M}_{\odot }$ for iPTF13asv. This shell increases the total mass of the system to ${1.59}_{-0.12}^{+0.45}{M}_{\odot }$ The ${}^{56}\mathrm{Ni}$, shell, and total masses of iPTF13asv are similar to those derived for SN20080522-000 in Scalzo et al. (2012).

The detached shell has little effect on the γ-ray opacity and peak luminosity of an SN. Since the rise time is proportional to ${M}_{\mathrm{tot}}^{1/2}$, the massless detached shell will not make the SN rise substantially longer than usual.

Strong UV emission in an SN Ia may be powered by an extrinsic SN-companion collision (Kasen 2010). In fact, in the UV-luminous SN2011de, Brown (2014) offered a possible explanation for its Swift light curve as a collision between the SN ejecta and a companion star. Here we consider the same model to interpret the strong UV emission of iPTF13asv.

We utilize the scaling relation in Kasen (2010) to fit the observed uvm2 light curve. In order to account for the non-negligible emission from the SN itself, we use the well-sampled uvm2 light curve of SN2011fe as a template. The fitting result shows that the companion star is located at $2\times {10}^{13}\;{\rm{cm}}$ away from the exploding WD (left panel of Figure 14). Given a typical mass ratio of a few, the companion has a radius of $\sim 100{R}_{\odot }$ and fills its Roche lobe.

Zoom In Zoom Out Reset image size

Although the model fit to the UV light curve looks plausible, it overpredicts the SN emission at very early phases. In the R-band light curve within a few days of explosion, the model-predicted SN flux ($200\;\mu \mathrm{Jy}$) is higher than the observed fluxes (the inset of Figure 1) by a factor of $\gt 30\%$. At −13.7 days, the predicted thermal emission flux from the model below 4000 Å is also much higher than the observed spectrum (right panel of Figure 14). Hence, we conclude that the strong UV emission seen in iPTF13asv is not produced by SN-companion collision.

As a result of the above analysis, we are forced to conclude that the strong UV emission is intrinsic. In fact, the strong UV emission and the lack of iron in early-phase spectra are probably causally related, as the iron group elements are the major absorbers of UV photons. These two observational facts, together with the near-IR secondary peak, strongly suggest that iPTF13asv has a stratified ejecta along the line of sight, with a strong concentration of iron group elements near the center of the explosion.

In Table 4, we summarize a comparison of normal SNe, iPTF13asv, and super-Chandrasekhar SNe. As can be seen from the table, on the one hand, iPTF13asv shares similar light curve shapes and a near-IR secondary peak with normal events. SNID also finds decent spectral matches between iPTF13asv and normal events. On the other hand, the peak radiation of iPTF13asv is as bright as super-Chandrasekhar events in both optical and UV. The evolution of Si ii velocities of iPTF13asv is also similar to those of super-Chandrasekhar events. In addition, we derived an total ejecta mass slightly beyond the Chandrasekhar mass limit. Hence, we classify iPTF13asv as an intermediate case between normal and super-Chandrasekhar subclasses.

Table 4.  Comparison of iPTF13asv to Normal and Super-Chandrasekhar SNe

Feature	Normal	iPTF13asv	Super-Chandrasekhar	Sectiona
B-band absolute peak magnitude	−18 to −19.5	−19.97 ± 0.06	$\lt -19.6$	Section 3.3
UV absolute peak magnitude	$\gtrsim -15$	−15.25	$\lesssim -15$	Section 3.3 and Figure 7
B-band ${\rm{\Delta }}{m}_{15}$ (mag)	0.8–1.2	1.0	$\simeq 0.6$	Section 3.3
Near-IR secondary peak	strong	strong	weak or absent	Section 3.3 and Figure 6
SNID	normal	normal	super-Chandrasekhar	Section 3.4.2 and Table 3
Carbon feature after max	no	weak	strong	Section 3.4.4 and Figures 10, 11
Si ii 6355 velocity at max (${10}^{3}\;{\rm{km}}\;{{\rm{s}}}^{-1}$)b	10–14	10	8–12	Section 3.4.3 and Figure 8
Si ii 6355 velocity gradient at max (${\rm{km}}\;{{\rm{s}}}^{-1}\;{{\rm{day}}}^{-1}$)b	50–150	∼0	−72–46	Section 3.4.3 and Figure 9
56Ni mass (${M}_{\odot }$)c	0.3–0.6	${0.81}_{-0.18}^{+0.10}$	$\gt 0.75$	Section 4.2
Ejecta mass (${M}_{\odot }$)c	0.8–1.5	${1.59}_{-0.12}^{+0.45}$	$\gt 1.5$	Section 4.2

Notes.

^aThis column points to sections in this paper that discuss corresponding features. ^bThe velocity measurements of normal events are from Foley et al. (2011). Those of super-Chandrasekhar events are from Scalzo et al. (2012). ^cThe mass measurements of normal events are from Scalzo et al. (2014c). Those of super-Chandrasekhar events are from Scalzo et al. (2012).

Download table as:  ASCIITypeset image

In addition to the features listed in the table, the H-band break, a sharp spectral feature formed by absorption of Fe ii, Co ii, and Ni ii (Hsiao et al. 2013), is also distinctive between super-Chandrasekhar and normal events. The H-band break emerges around the maximum for normal SNe and decays to disappear within a month of maximum. In contrast, this feature does not appear in the super-Chandrasekhar events. However, the only near-IR spectrum of iPTF13asv is taken one month after the maximum. Therefore, we cannot determine whether iPTF13asv shows the H-band break or not.

The massive ejecta and the stratification of the ejecta favor a DD progenitor system for iPTF13asv. In an SD system, a non-rotating WD cannot exceed the Chandrasekhar mass limit, and it is not clear in reality how rotation could increase this mass limit. Hydrodynamic simulations (e.g., Kromer et al. 2010; Sim et al. 2013) also show that explosions in an SD system cannot avoid a certain level of mixing in the ejecta. Hence, these models do not easily concentrate most of the iron group elements in the center of the ejecta. For merging WDs, in contrast, simulations of prompt detonation (e.g., Moll et al. 2014) produce strongly stratified structures along polar directions in asymmetric ejecta. In these directions, iron group elements are confined to the low-velocity regions.

We also consider the core-degenerate scenario to explain the progenitor system of iPTF13asv (e.g., Soker et al. 2014), but more exploration in this scenario is needed to explain the weak mixing of iron group elements in the fast-moving ejecta, persistent carbon features after maximum, and low but almost constant Si ii velocity.

Unsurprisingly, iPTF13asv is an outlier from the Phillips relation (Phillips 1993). As shown in Figure 15, iPTF13asv is above the empirical relation by half a magnitude.

Zoom In Zoom Out Reset image size

To have better calibration in cosmology, a third color term is introduced in the Phillips relation, i.e.,

$$\begin{eqnarray}&&\mu ={m}_{B}^{*}-({M}_{B}-\alpha {x}_{1}+\beta c),\end{eqnarray} \tag{ 6 }$$

where μ is the distance modulus; ${m}_{B}^{*}$ is the observed peak magnitude in the rest-frame B band; and α, β and M_B are free parameters. To account for the dependence on the host galaxy properties, Sullivan et al. (2011) suggest to using different values of M_B for galaxies of stellar mass greater than and less than ${10}^{10}{M}_{\odot }$. In the case of iPTF13asv, the stellar mass of its host galaxy is $\sim {10}^{7.8}{M}_{\odot }$. For galaxies with stellar mass less than ${10}^{10}{M}_{\odot }$, Betoule et al. (2014) used a fiducial value of ${H}_{0}=70\;{\rm{km}}\;{{\rm{s}}}^{-1}\;{{\rm{Mpc}}}^{-1}$ and obtained ${M}_{B}=-19.04\pm 0.01$, $\alpha =0.141\pm 0.006$, and $\beta =3.101\pm 0.075$. Using the same H₀ and the iPTF13asv measurements of ${m}_{B}^{*}=16.28\pm 0.03$, ${x}_{1}=0.37\pm 0.09$, and $c=-0.16\pm 0.02$, we find that iPTF13asv can still be included in this empirical relation and thus be useful for cosmographic measurements, whereas super-Chandrasekhar events are outliers of this empirical relation (Scalzo et al. 2012).

Spectroscopic classification for high-redshift SNe requires very long integration on big telescopes. Therefore, in high-redshift SN surveys, an optical–UV "dropout" is introduced to preselect type Ia candidates. For example, Riess et al. (2004, 2007) used the F850LP, F775W, and F606 filters on the HST Advanced Camera for Surveys (ACS) to search for SNe at redshifts up to 1.8.

The color preselection criteria may introduce bias by ignoring UV-luminous SNe. In Figure 16, we calculate the color difference in the F850LP, F775W, and F606 filters for normal SN1992a (Kirshner et al. 1993), near-UV blue SN2011fe (Mazzali et al. 2014), and UV-luminous iPTF13asv at different redshifts. As can be seen, the three SNe show different colors at high redshifts. Although the rate of iPTF13asv-like events is probably low in the nearby universe, there might be more such SNe at high redshifts as there are more metal-poor dwarf galaxies at high redshifts. Hence it might become a non-negligible component in estimating the SN rate at high redshifts.

Zoom In Zoom Out Reset image size